Refactored prt of the parser implementation (#159)

Author: glynos

CMakeLists.txt

@@ -4,7 +4,7 @@
 # http://www.boost.org/LICENSE_1_0.txt)
 
 
-cmake_minimum_required(VERSION 3.17)
+cmake_minimum_required(VERSION 3.16)
 
 project(
         skyr-url

README.md

@@ -139,7 +139,7 @@ int main() {
 
   std::cout << "Domain?   " << std::boolalpha << url.is_domain() << std::endl;
   std::cout << "Domain:   " << url.hostname() << std::endl;
-  std::cout << "Domain:   " << url.u8domain() << std::endl;
+  std::cout << "Domain:   " << url.u8domain().value() << std::endl;
 
   std::cout << "Port:     " << url.port<std::uint16_t>().value() << std::endl;
 
@@ -161,6 +161,8 @@ Here is the ``CMake`` script to build the example:
 ```cmake
 # CMakeLists.txt
 
+cmake_minimum_required(VERSION 3.16)
+
 project(my_project)
 
 find_package(tl-expected CONFIG REQUIRED)

include/skyr/v2/core/url_parser_context.hpp

@@ -520,35 +520,30 @@ class url_parser_context {
 
   auto parse_file(char byte) -> tl::expected<url_parse_action, url_parse_errc> {
     set_file_scheme();
+    set_empty_host();
 
     if ((byte == '/') || (byte == '\\')) {
       if (byte == '\\') {
         *validation_error |= true;
       }
       state = url_parse_state::file_slash;
     } else if (base && (base->scheme == "file")) {
-      if (is_eof()) {
-        set_host_from_base();
-        set_path_from_base();
-        set_query_from_base();
-      } else if (byte == '?') {
-        set_host_from_base();
-        set_path_from_base();
+      set_host_from_base();
+      set_path_from_base();
+      set_query_from_base();
+      if (byte == '?') {
         set_empty_query();
         state = url_parse_state::query;
       } else if (byte == '#') {
-        set_host_from_base();
-        set_path_from_base();
-        set_query_from_base();
         set_empty_fragment();
         state = url_parse_state::fragment;
       } else {
+        clear_query();
         if (!details::is_windows_drive_letter(still_to_process())) {
-          set_host_from_base();
-          set_path_from_base();
           details::shorten_path(url.scheme, url.path);
         } else {
           *validation_error |= true;
+          clear_path();
         }
         state = url_parse_state::path;
         if (input_it == std::begin(input)) {
@@ -574,11 +569,11 @@ class url_parser_context {
       }
       state = url_parse_state::file_host;
     } else {
-      if (base && ((base->scheme == "file") && !details::is_windows_drive_letter(still_to_process()))) {
-        if (!base->path.empty() && details::is_windows_drive_letter(base->path[0])) {
+      if (base && (base->scheme == "file")) {
+        set_host_from_base();
+        if (!details::is_windows_drive_letter(still_to_process()) &&
+            (!base->path.empty() && details::is_windows_drive_letter(base->path[0]))) {
           set_path_from_base0();
-        } else {
-          set_host_from_base();
         }
       }
 
@@ -849,6 +844,10 @@ class url_parser_context {
     url.cannot_be_a_base_url = true;
   }
 
+  void clear_path() {
+    url.path.clear();
+  }
+
   void add_empty_path_element() {
     url.path.emplace_back();
   }
@@ -877,6 +876,10 @@ class url_parser_context {
     url.path[0] += pct_encoded.to_string();
   }
 
+  void clear_query() {
+    url.query->clear();
+  }
+
   void set_empty_query() {
     url.query = std::string();
   }

src/v1/core/url_parser_context.cpp

@@ -465,35 +465,30 @@ auto url_parser_context::parse_port(char byte) -> tl::expected<url_parse_action,
 
 auto url_parser_context::parse_file(char byte) -> tl::expected<url_parse_action, url_parse_errc> {
   url.scheme = "file";
+  url.host = host{empty_host{}};
 
   if ((byte == '/') || (byte == '\\')) {
     if (byte == '\\') {
       *validation_error |= true;
     }
     state = url_parse_state::file_slash;
   } else if (base && (base->scheme == "file")) {
-    if (is_eof()) {
-      url.host = base->host;
-      url.path = base->path;
-      url.query = base->query;
-    } else if (byte == '?') {
-      url.host = base->host;
-      url.path = base->path;
+    url.host = base->host;
+    url.path = base->path;
+    url.query = base->query;
+    if (byte == '?') {
       url.query = std::string();
       state = url_parse_state::query;
     } else if (byte == '#') {
-      url.host = base->host;
-      url.path = base->path;
-      url.query = base->query;
       url.fragment = std::string();
       state = url_parse_state::fragment;
     } else {
+      url.query = std::nullopt;
       if (!is_windows_drive_letter(input.substr(std::distance(begin(input), it)))) {
-        url.host = base->host;
-        url.path = base->path;
         shorten_path(url.scheme, url.path);
       } else {
         *validation_error |= true;
+        url.path.clear();
       }
       state = url_parse_state::path;
       if (it == begin(input)) {
@@ -519,10 +514,11 @@ auto url_parser_context::parse_file_slash(char byte) -> tl::expected<url_parse_a
     }
     state = url_parse_state::file_host;
   } else {
-    auto substr = input.substr(std::distance(begin(input), it));
-    if (base && ((base->scheme == "file") && !is_windows_drive_letter(substr))) {
-      if (!base->path.empty() && is_windows_drive_letter(base->path[0])) {
-        url.path.push_back(base->path[0]);
+    if (base && (base->scheme == "file")) {
+      auto substr = input.substr(std::distance(begin(input), it));
+      if (!is_windows_drive_letter(substr) &&
+          (!base->path.empty() && is_windows_drive_letter(base->path[0]))) {
+          url.path.push_back(base->path[0]);
       } else {
         url.host = base->host;
       }